Tumors progress through the continuous accumulation of genetic and epigenetic changes that enable escape from normal cellular and environmental controls. These aberrations may involve genes that affect cell-cycle control, apoptosis, angiogenesis, adhesion, transmembrane signaling, DNA repair, and genomic stability. A number of genes that contribute to this process have already been discovered. However, large-scale analysis of gene expression and gene copy number suggest that the number of such genes may be large, perhaps strikingly so, and many important cancer-related genes remain to be discovered. Identification of recurrent changes in gene copy number, organization, sequence or expression is one common approach to identification of genes that play a role in cancer. Large-scale array analysis techniques for assessment of genome copy number, expression level and DNA sequence polymorphisms are now accelerating the rate at which tumors can be analyzed. These same technologies are promising as diagnostic platforms that can be employed to assess specific changes in individual tumors thereby permitting selection of therapeutic strategies that are optimal for these tumors.
Array based comparative genomic hybridization (CGH), allows the changes in relative DNA sequence copy number to be mapped onto arrays of cloned probes. In array CGH, total genome DNAs from tumor and reference samples are independently labeled with different fluorochromes or haptens and co-hybridized to normal chromosome preparations along with excess unlabeled Cot-1 DNA to inhibit hybridization of labeled repeated sequences. The principle advantages of CGH are that it maps changes in copy number throughout a complex genome onto a normal reference genome so the aberrations can be easily related to existing physical maps, genes and genomic DNA sequence. In addition, array CGH allows quantitative assessment of DNA sequence dosage from one copy per test genome to hundreds of copies per genome. Initial work involved CGH to arrays comprised of targets spanning >100 kb of genomic sequence, such as BACs. More recently, CGH to cDNA arrays has been demonstrated. cDNA arrays are attractive for CGH since they are increasingly available and carry a very large number of clones. In addition, the same array can be used to assess expression and copy number.
Single nucleotide polymorphisms (SNPs) also can be detected efficiently by hybridization of fluorescently labeled PCR amplified representations of the genome to arrays comprised of oligonucleotides. Both alleles of each of several thousand SNP markers and single-base mismatch targets may be presented on an array. The stringency of the hybridization reaction is adjusted so that hybridization is diminished if a single base mismatch exists between the probe and oligonucleotide substrate. Thus, its hybridization signature can determine the presence or absence of an allele in the hybridization mixture. This technique is rapid and scales well to genome-wide assessments of linkage or LOH (loss of homogeneity).
Enormous progress has been made in recent years in the development and DNA sequence characterization of cDNA clones from the human, mouse and other model organisms. In humans, these data have been computationally assembled into over 8000 genes and 83,000 clusters. The cDNA clones associated with these sequences are publicly available. These clones and their associated sequences form the basis for a powerful microarray approach to large-scale analysis of gene expression. In this approach, labeled mRNA samples are hybridized to arrays of cDNA clones or oligonucleotides derived from the associated sequences. The arrays may be on silicon or membrane substrates. The labeled probes may be labeled radioactively or with fluorescent reagents so that the resulting hybridization signals can be detected using autoradiography, phosphoimaging or fluorescence imaging.
cDNA and oligonucleotides arrays have been made using robots to move DNA from microtiter trays to silicon substrates or to nylon membranes. This approach is flexible and is especially well-suited to production of custom arrays, but also has been applied to make large-scale arrays carrying 40,000 different clones. An alternative is to synthesize oligonucleotide arrays directly on silicon substrates using photolithographic approaches. These techniques work by projecting light through a photolithographic mask onto the synthesis substrate. Single oligonucleotide arrays on silicon substrates have been constructed with elements representing more than 40,000 genes/ESTs.
The conventional array approaches described above, while powerful, are limited by the inefficient manner in which probe is used and by the long hybridization times required. These limitations arise from the need to distribute probe molecules over a large surface during hybridization. As a result, most probe molecules are far from the targets to which they might hybridize and sensitivity suffers. This reduces the rate at which hybridization occurs and results in most probe molecules never reaching the targets to which they might bind, a phenomenon that becomes increasingly limiting for long oligonucleotides with slow diffusion rates. This problem can be reduced by using relatively large amounts of probe, vigorous mixing and using space-filling reagents such as dextran sulfate in the hybridization mixtures. However, substantial improvement is still needed to allow practical use of DNA or RNA recovered from small amounts of material (e.g., collected by microdissection) and to increase the rate of hybridization.